42 results about "Random effects model" patented technology

The invention discloses a short period load prediction method based on kernel principle component analysis and a random forest. The a short period load prediction method comprises the following steps of: (1) analyzing and selecting data influencing load prediction precision of a day to be predicted in an operational electric power system, and preliminarily constructing training and prediction sample sets; (2) utilizing kernel principle component analysis to carry out dimensionality reduction on training sample data; (3) utilizing a random forest model to train the training sample data after the dimensionality reduction, and obtaining the random forest model after the training; and (4) inputting prediction sample data into the random forest model after the training, and carrying out short period load prediction of the day to be predicted. The short period load prediction method has the advantages that the kernel principle component analysis and the random forest model are combined for carrying out short period load prediction on the electric power system, the prediction precision, efficiency and data rationality are improved.

The invention provides a method and a device for identifying abnormal users based on a random forest model, belonging to the technical field of big data. The method comprises the following steps: thesample data is counted from the information of a history user according to a preset attribute, wherein, the preset attribute comprises a first class attribute and a second class attribute, and the classification tag of the history user is obtained; a random forest model is trained using the sample data and the classification label, wherein in the training process, the first class of attributes corresponds to a first sampling probability, the second class of attributes corresponds to a second sampling probability, and the first sampling probability is greater than the second sampling probability; According to the preset attribute, the target data is counted from the information of the user to be identified, and the target data is processed through the trained random forest model to determine whether the user to be identified is an abnormal user. The present disclosure may reduce the amount of sample data required for the abnormal user identification method and improve the accuracy of identification.

The invention relates to a differential evolution random forecast classifier-based photovoltaic array fault diagnosis method. The method comprises the steps of firstly, collecting photovoltaic array voltages under various working conditions and currents of photovoltaic strings, and performing identification on various working conditions by different identifiers; secondly, determining a quantity range of decision trees in a random forest model by adopting an out-of-bag data-based classification misjudgment rate mean value; thirdly, performing global optimization on the quantity range of the decision trees by utilizing a differential evolution algorithm to obtain an optimal decision tree quantity value; fourthly, substituting the calculated optimal decision tree quantity value into a randomforecast classifier, and training samples to obtain a random forecast fault diagnosis training model; and finally, performing fault detection and classification on a photovoltaic array by utilizing the training model. According to the method, the model training speed can be greatly increased while the optimal model classification accuracy is ensured, so that the fault detection and classificationof the photovoltaic power generation array are realized more quickly and accurately.

The invention discloses a transformer fault diagnosis method for optimizing a random forest model based on a particle swarm algorithm. The method comprises the steps: dividing a training set and a test set by taking a non-coding ratio of analysis data of dissolved gas in transformer oil as characteristic vector input; constructing a random forest model, and optimizing the random forest model through a particle swarm optimization algorithm to obtain two optimal parameters; and rebuilding a random forest model according to the obtained optimal parameters to identify the fault type of the transformer. According to the method, the fault diagnosis accuracy of the transformer is effectively improved, and a reliable basis is provided for operation and maintenance personnel to correctly judge theoperation condition of the transformer.

Provided is a pipeline health state assessment method based on a random forest model, belonging to the city water supply pipe network technical field. The method comprises: respectively extracting the pipeline basic information and historic damage case from the foundation database and the damaged database of a city water supply network; performing data pre-processing on obtained pipeline information; utilizing a random forest model to establish the relation between an independent variable and a dependent variable, and evaluating model classification effects; utilizing the random forest model passing classification effect evaluation to predict the damage probability of the water supply pipe network; grading prediction results, representing health levels by different colors, and drafting a health state thematic map; and evaluating pipeline damage influence factor importance, and analyzing influence rules. The pipeline health state assessment method can assess pipe network health states, obtain prediction results fundamentally according with real conditions, effectively evaluate pipeline states, and provide certain theoretical support for water supply enterprises to make a plane of determining pipeline maintenance and reconstruction priorities, and optimizing maintenance.

The invention discloses a random forest model-based population data spatialization method. According to the method, population distribution-related variable factors such as surface coverage data and lamp light data are selected; the population distribution-related variable factors are pre-processed, and the pre-processed population distribution-related variable factors are inputted into a random forest model; the relationship between population density and the variable factors, and the importance of the variable factors are determined through using the random forest model; the population density of each grid is obtained through inversion based on the relationship; and an estimation result is corrected through regional density charting, so that a gridded population distribution result can be obtained. With the method adopted, the precision of population data spatialization can be further improved, and the importance of the variable factors are interpreted.

The invention discloses a driving fatigue detection system and identification method based on eye movement index data. A data acquisition module collects four kinds of eye movement data according to a frequency of 200HZ; a data analysis processing module transmits the eye movement data into a computer program to carry out preprocessing; the eye movement index data of a driver and a driver fatigue degree corresponding to each data are collected to be original data to establish a random forest model, the classification judgment of the eye movement data of the driver is carried out through each decision tree in the random forest model, and a result is outputted; finally the classification results of decision trees are integrated, voting is carried out by using the random forest model, and a result with a highest integrated voting probability is the final result of the classification. The random forest model of machine learning is employed, the training speed is fast, the classification and prediction performance is excellent, whether the driver is fatigue at present can be identified quickly, the random forest model is continuously increased and updated with data amount, and the discrimination performance is continuously optimized and improved.

The invention relates to a gene classification method and system based on clustering and random forest algorithms and belongs to the technical field of biological information. The method comprises a step of acquiring gene sample data, clustering the acquired gene sample data by using the clustering algorithm to obtain a cluster center, and supplementing a training sample set with an obtained cluster center set, a step of adjusting the number of fixed decision tree random description attributes in a traditional random forest algorithm to a random value, wherein on one hand, strong decision trees in a decision tree set are kept, on the other hand, the number of average random description attributes of the decision tree set is reduced, thus the correlation between the decision trees is further reduced, and a step of predicting genetic data to be classified by using each decision tree in a random forest model. According to the method and the system, the cluster center obtained through theclustering algorithm is taken as artificial data to expand the training set of the random forest model, thus the random forest model is fully trained, the obtained classification model has high precision, and the accuracy of the classification of genetic data is improved.

The invention discloses a sequence characteristic analysis method for forecasting a miRNA target gene. The method comprises the steps of constructing related characteristics of 27 miRNA-target point pairing sequences on the basis of a CLASH experiment data set, and forming a characteristic set comprising 84 characteristic values by combining traditional characteristic; and performing machine learning by using a random forest model, and constructing a miRNA target gene forecast model to perform miRNA target gene recognition. The model constructed according to the method has the advantages of high accuracy, sensitivity, specificity and precision, and the miRNA target gene can be relatively and accurately forecasted.

The invention relates to a pedestrian recognition system based on a random forest support vector machine. The pedestrian recognition system comprises a characteristic extraction module, a clustering module, a random forest creating module and a scoring model module. The invention also relates to a pedestrian recognition processing method based on the random forest support vector machine. A similarity ranking way is used for replacing the comparison of traditional similarity absolute values, a threshold value does not need to be delimited, and an obtained ranking result is convenient for users to judge; and since multiple characteristics are required for establishing a random forest model and samples can not be subjected to mutual classification perfection only from apparent characteristics, a K-means clustering algorithm is adopted to replace a phenomenon that a sample category is manually given, and potential relationships among samples can be mined. The method and the system exhibit robustness on pedestrian posture change and can eliminate interferences from other types of samples when the similarity is calculated, a ranking result of RankSVM (Support Vector Machine) is in the top, and recognition accuracy can be improved when the similarity is calculated. Compared with traditional algorithms including MCC, RankSVM and the like listed in the prior art, the pedestrian recognition system is high in recognition accuracy.

The invention discloses a transformer fault diagnosis method based on a random forest. The method comprises the following steps of collecting fault gas concentration data in insulating oil in a transformer and a corresponding fault type as a training sample; according to the training sample, based on the generation steps of a decision tree, establishing a fault decision tree; according to the fault decision tree, establishing a random forest model; and collecting the fault gas concentration data of a unknown fault type, inputting into the random forest model so as to acquire the fault type through the random forest model. The fault gas concentration data in the insulating oil in the transformer is taken as the training sample so as to establish the random forest model, a whole transformerfault can be accurately diagnosed, stability is high, and the method can be applied to the transformer diagnosis technology field.

The invention provides a power industrial control attack classification method and system based on machine learning. The method and the system are characterized by utilizing the historical message data of the electric power industrial control, after completing the default value of the data, extracting the characteristic variable, inputting the stochastic forest model for multi-fold cross-validation, and adjusting the model parameters according to whether the stochastic forest model has occurred fitting and / or under-fitting phenomenon to determine the optimal stochastic forest model to classifythe electric power industrial control attacks. Compared with that prior art, by collecting the history message data of electric power industry control for machine learning, the random forest model isbuilt, and the messages generated by the electric power industrial control system are imported into the random forest model to realize the classification of the electric power industrial control attacks, thereby improving the status quo of the passive defense of the industrial control system, enabling the system to detect and intercept the attacks before being attacked, and improving the safety performance of the electric power industrial control system.

The invention relates to a method for recognizing Raman spectrum substances. The method includes S100, selecting a plurality of samples, generating Raman spectrogram data sets of the samples, and preprocessing all Raman spectrograms in the Raman spectrogram datasets to automatically eliminate factors with influence on the spectrogram analysis accuracy; S200, extracting sample features from all thepreprocessed Raman spectrograms; S300, building a plurality of random forest models according to the Raman spectrogram data sets and the extracted sample features; S400, selecting the optimal randomforest models from the multiple random forest models and judging target substance categories by the aid of the optimal random forest models. The sample features are feature vectors applicable to the random forest models. The to-be-detected samples belong to the target substance categories. The method has the advantages that Raman spectrum substance recognition (qualitative analysis) problems are converted into machine learning classification problems, batch real-time processing can be implemented, and accordingly the operating speeds can be greatly increased on the basis that the high accuracyis guaranteed.

The invention provides a positioning method based on a random forest model. The method comprises the steps that the equipment signal data of multiple sets of wireless equipment reported by a mobile terminal are received; the signal intensity data of each set of wireless equipment are extracted from the equipment signal data to act as the eigenvalue, and the eigenvalue of each set of wireless equipment is combined so as to generate a feature vector; the feature vector is inputted to a pre-determined indoor positioning classifier, and multiple possible position coordinates of the mobile terminaland the probability of each possible position coordinates are calculated; and the position coordinate of the highest probability are selected from the prediction result outputted by the indoor positioning classifier and the position coordinate of the highest probability are transmitted to the mobile terminal. The invention also provides an electronic device and a computer readable storage medium.With application of the positioning method based on the random forest model, the device and the storage medium, the positioning accuracy of the mobile terminal can be enhanced.

The invention relates to the field of medical image processing and application, in particular to a Chinese brain language area distribution graph construction method. The invention provides a new group-based reliable Chinese language distribution graph construction method for a current situation of locating of a Chinese brain language area not universally accepted yet in the prior art. According to the method, a Chinese brain language area distribution graph is constructed based on intraoperative cortical electrical stimulation and can be applied to the neurosurgery to accurately locate the Chinese brain language area. The construction method mainly comprises the steps of constructing a two-dimensional probability graph, a three-dimensional surface distribution graph, and a random effect model-based statistic parameter graph; and displaying the distribution situation of the language function area in the brain from two-dimensional and three-dimensional perspectives. According to the method, the deviation caused by operation of locating the Chinese language function area only by MRI can be effectively avoided; the Chinese language area can be accurately located through intraoperative language locating; and better reference and guidance are provided for clinical surgeries.

The invention discloses a DNS protocol covert channel detection method based on a random forest. According to the method, a random forest algorithm in machine learning is used to learn features; according to the method, malicious traffic of a DNS hidden channel can be quickly identified, the DNS hidden channel can be effectively detected, DNS requests and response traffic are analyzed by analyzingthe DNS traffic, feature extraction is performed on common fields for transmitting hidden information, and statistics of normal threshold ranges is performed on features; and then learning normal andabnormal DNS traffic characteristics by using a random forest model, establishing a model, and identifying whether the traffic is abnormal traffic, thereby realizing detection of a hidden channel.

The present invention provides a flight arrival time prediction method. The method comprises the steps of: obtaining historical data, performing preprocessing of the historical data, and obtaining characteristic data, wherein the characteristic data comprises weather characteristic data, flight historical flight characteristic data and position characteristic data; taking the characteristic data and actual time of arrival of one flight corresponding to each characteristic data as original data to train a random forest model; performing classification of the characteristic data of a flight to be predicted by employing the random forest model, and performing voting of results obtained through classification, wherein a result having the most votes is predicted arrival time of the flight to bepredicted. According to the invention, the problem is solved that predicted arrival time of flights is not accurate, and the accuracy of the predicted arrival time of the flight is improved.

The invention discloses an electricity market monthly electricity utilization prediction method. The prediction method comprises the following steps of step 1, setting an electricity utilization prediction model, defining a predicted state space model, performing complementary prediction by using a random forest model, loading randomForest, rpart software packages into R software, and leading in an explaining variable and an explained variable; and step 2, determining a prediction model input amount, establishing an air temperature aggregative indicator, adjusting a mobile holiday effect through an effective working day method, predicating a leading indicator, and determining a leading period through model measurement and calculation and coefficient calculation. According to the electricity market monthly electricity utilization prediction method, after the air temperature aggregative indicator, the leading indicator, the business expanding prediction indicator and the like are determined, a state space vector model and a random forest machine study model are combined for performing electricity consumption prediction, so that the prediction method is more accurate and effective.

The invention discloses a personal credit default prediction method which specifically comprises the following steps: S1, collecting personal information data of a borrower and credit account activityinformation data to establish a database, removing missing values and abnormal values of the data, and preprocessing the data; S2, constructing a decision-making tree, forming a random forest by a plurality of combined classifiers of the decision-making tree, constructing a random forest model, and classifying credit data samples; S3, logical regression analysis and a random forest algorithm arecombined, the classification accuracy is improved, and personal credit default prediction is completed. According to the method, the diversity of personal data is more focused, information acquisitionand analysis are comprehensive, and the defect that the prior art depends on data fitting is overcome; The method has the advantages of wide application range, small individual influence and high prediction accuracy.

The present invention proposes a real estate valuation method based on a random forest, comprising: step 1, establishing a basic database step: collecting real estate data to form a basic database, and using a clustering algorithm or / and a deduplication algorithm or / and a screening algorithm to perform data processing on the basic database , carry out characteristic variable quantitative processing to real estate data; Step 2 establishes the step of random forest model, step 3 steps the step that random forest model is tested, step 4 real estate valuation step, the random forest algorithm of the present invention does not need pre-set function form, can Accurately fit samples of complex nonlinear relationships.

The invention discloses an intranet transaction identification method based on a random forest and a naive Bayes model. The method comprises the following steps: acquiring an internal screen transaction sample data set under different time window periods, screening and constructing a characteristic index set by adopting a random forest model, constructing a Bayesian identification model of the internal screen transaction according to the screened characteristic index set, and performing internal screen transaction identification by adopting the Bayesian identification model to obtain a resultof whether the internal screen transaction exists or not; and after the event, supervising and verifying whether the internal transaction recognition result is correct, and training and updating the Bayesian recognition model according to the recognition result. According to the invention, the stock intranet transaction identification model is established, so that whether the test target is subjected to intranet transaction or not is accurately identified; a quasi-Newton method and a genetic algorithm are combined, so that parameters of the random forest model are quickly optimized to an optimal solution with high precision, and the solution of the optimal solution has small dependence on an initial value; the method is easy to implement and stable in performance, and robustness and accuracy can be further improved along with increase of sample data.

The invention is applicable to the technical field of data processing, and provides a method and a system for optimizing random forest models. The method includes creating heat distribution histograms of the random forest models and distribution histograms of decision trees, with different prediction accuracies, in the random forest models; computing similarity degrees among the decision trees by the aid of proportions of identical attribute nodes among the decision trees according to the heat distribution histograms and the distribution histograms of the decision trees, with the different prediction accuracies, in the random forest models; deleting the decision trees with the minimum prediction accuracies according to the distribution histograms of the decision trees, with the different prediction accuracies, in the random forest models, and / or deleting the decision trees with the highest similarity degrees among the decision trees in the random forest models according to the computed similarity degrees among the decision trees. The method and the system have the advantages that the random forest models optimized by the aid of the method and the system are small in scale and high in prediction accuracy and prediction speed, the prediction efficiency of the random forest models can be effectively improved, and the like.

The invention relates to a method for detecting suspected code plagiarism based on a random forest model. The method can be divided into two stages from a large aspect, the first stage is a feature extraction stage, and the second stage is a model training and prediction stage. The method collects the relevant characteristic data of the two codes to be detected, the characteristic data of the users to be detected and the attributes of the related topics, and modeling is carried out by introducing a random forest algorithm to obtain whether the current users are suspected of plagiarism or not.

The invention discloses a chromatographic overlapping peak analytical method based on wavelet transform and a random forest model. The chromatographic overlapping peak analytical method comprises thesteps that a plurality of chromatographic overlapping peak signals are generated in a simulation mode according to different parameters; for each of the overlapping peak signals, gaus1 wavelets are used for performing the wavelet transform to simulate a first-order derivative; a simulated first-order derivative curve is used for calculating and obtaining four curve inflection points of the original chromatographic overlapping peak signals; the curve inflection points are divided into a training set and a test set according to a certain proportion; the horizontal and vertical coordinates of thefour inflection points are used as the input, a sub-peak area ratio is used as the output, and the optimal parameter of a model is determined by using a cross-validation method in the training set; the random forest model is constructed and trained in a supervised mode according to the optimal parameter; and the test set is used for verifying the effect of the model; and the same method is used for detecting the inflection points of the actual overlapping peak signals, and the trained model is used for performing fitting calculation on the own sub-peak area ratio. The chromatographic overlapping peak analytical method improves the accuracy of the analytical result, and has the advantages of high model convergence speed, simple parameter adjustment and high training efficiency.

The invention discloses a polarized SAR image random forest classification method integrating multiple features. The method comprises the steps that an SLIC superpixel generation algorithm is utilizedto segment a to-be-classified polarized SAR image; feature information of the polarized SAR image is extracted, and a high-dimensional polarized feature map is constructed; a random forest model is trained based on the high-dimensional polarized feature map, and a polarized SAR image random forest model is constructed; the polarized SAR image random forest model is utilized to perform statisticalanalysis on the number of votes for each category of pixels, and a superpixel category probability graph based on a random forest is constructed with the superpixels being units; each superpixel category probability is iteratively corrected based on a PLR model, and a superpixel category probability graph after iterative correction is obtained; and each superpixel category is calculated with thesuperpixels being units, and a classification result is output. According to the method, the improved SLIC algorithm is utilized to generate accurate and fine superpixels as classification units, so that interference of speckle noise in the polarized SAR image is effectively lowered; and by use of neighborhood features of the superpixels, the interference of the speckle noise is further reduced, and the precision of the classification result is improved.

The invention discloses a key information infrastructure asset identification method combined with a mixed random forest, and belongs to the technical field of computers and information science. The method comprises the following steps of carrying out the structured processing on the collected facility asset data and carrying out the feature optimization expression to obtain an extended feature vector; in combination with a Delphi expert consultation method and a principal component analysis method, analyzing the key influence factors of the asset facilities, and extracting the key feature vectors; combining the plurality of random forest judgment models with a gating function to obtain a mixed random forest judgment model; and based on the constructed mixed random forest model, identifying whether the traffic is a key asset infrastructure or not. According to the key information infrastructure asset recognition method provided by the invention, the asset feature construction and the key factor extraction are realized by combining a machine learning method under big data, and the respective expert models are constructed by partitions, so that the recognition accuracy and efficiency are improved, and the generalization ability and expandability of the model are improved.

The invention discloses a user authentication password security evaluation method and device based on a random forest model. The user authentication password security evaluation device comprises a prefix feature extraction module, a training set reading and processing module, a model training module and a password generation module. The user authentication password security evaluation method includes the steps: improving the Markov model, taking each character of the password in the password training set as a category, extracting a prefix feature of the character as a feature vector, and training by adopting a random forest to obtain a probability model of a multi-classification problem; and for any character string, obtaining probability distribution of suffix characters of the prefix through the probability model, and generating a candidate password, thereby realizing security evaluation on the user password. According to the user authentication password security evaluation method and device, the problem that an original Markov model is prone to over-fitting due to the model fitting principle can be solved, and the attack effect is better, and the algorithm effect is more stable,and the password security can be evaluated more accurately.

The invention provides a power transmission and transformation suspicious data screening method and device based on a random forest model, and the method comprises the steps: S1, selecting data of multiple dimensions according to the category and periodicity rule of power transmission and transformation equipment, and building a data feature item; S2, distributing different weights for the data according to the sampling time, respectively marking known normal data and abnormal data as positive and negative samples, and dividing a data set into K parts; S3, training a random forest model by adopting a K-fold cross validation method, iteratively adjusting the number T of trees in the random forest by taking an average value of positive and negative sample accuracy as a target, and obtaininga value of T when an index is optimal; and S4, screening suspicious data by using the trained model. Power transmission and transformation equipment is used as an object to construct a suspicious datascreening object, a random forest model of an optimization training set is used for learning data rules from a large amount of historical sampling data, power transmission and transformation suspicious data identification and screening are achieved, the workload of manual screening is reduced, and the data quality of an electric power regulation and control system is improved.

Training a random forest model to relate settings of a network security device to undesirable behavior of the network security device is provided. A determination of a corresponding set of settings associated with each region of lowest incident probability is made using a random forest. The plurality of identified desired settings are presented as options for changing the network security device from the as-is settings to the identified desired settings. A choice is received from the plurality of options. The choice informs the random forest model. The random forest model ranks for a new problematic network security device the plurality of options for changing the new problematic network security device from as-is settings to desired settings by aggregating an identified cost of individual configuration changes, thereby identifying a most cost-effective setting for the network security device to achieve a desired output of the network security device.

The invention provides a training method, a wheat recognition method and a training system based on a random forest model, and relates to the technical field of computer application technology, and the method comprises the steps: obtaining the random forest model which comprises a plurality of decision trees; obtaining a plurality of pieces of training sample data, wherein the training sample datacomprises category data and feature sample data; sequentially inputting the plurality of training sample data into the random forest model to obtain a plurality of pieces of prediction data; calculating the precision of the random forest model according to the prediction data and the category data; And optimizing the random forest model according to the precision to obtain an optimized random forest model, thereby alleviating the technical problem of low ground object recognition efficiency by using a remote sensing technology.